Attributes in a contact center typically reflect properties and/or types of customers and work items in the contact center. Historically, a contact center has been able to determine what attribute areas have significant growth and what attribute areas have decayed by looking at volume and other indicators. Unfortunately, the job of monitoring all of the attributes and how they are staffed and how regularly used falls to the contact center administrator. This can be a significant job if the contact center is complex and distributed. It would be advantageous to have a graph-based contact center that could manage its own attributes within parameters set by the administrator.